Although autonomous driving technology has become an emerging research focus, safety is still the most crucial concern when autonomous vehicles leave research laboratory and enter public traffic. Direct yaw moment control (DYC), which differentially brakes the wheels to produce a yaw moment, is an important system to ensure the driving stability of vehicle under extreme conditions. The design of traditional DYC system must need to take into account driver’s intention and vehicle dynamics. However, for autonomous vehicle, no human is involved in driving process, and enforcing traditional DYC system may conﬂict with the demands of the desired path. Therefore, in this paper, a novel DYC system for autonomous vehicle is proposed using a hierarchical control architecture. In the yaw moment control layer, to ensure vehicle stability and suppress lateral path tracking deviation, a robust H∞ control strategy based on kinematics and dynamics of vehicle system is developed through linear matrix inequality (LMI). In the braking torque allocation layer, selection logic of the controlled wheel is designed, and desired braking control torque at the wheel is obtained. In the executive layer, the control scheme of integrated-electro-hydraulic brake (IEHB) actuator is developed to guarantee that the suitable ground braking force is generated. Finally, Matlab/Simulink-AMESim co-simulation test is carried out on a 7-DOF nonlinear vehicle model with the IEHB actuator for a double lane change maneuver. The results show that the proposed DYC system can effectively maintain autonomous vehicle stability while suppressing lateral path tracking error in dynamic driving situations at handling limits.